Skip to content

Browse by Topic¶

All 107 papers grouped by 57 topic tags. A paper appears under each of its tags. Generated by scripts/build_manifest.py - do not edit by hand.

Jump to: agents · alignment · architecture · attention · audio · benchmarks · chain-of-thought · code · computer-vision · controllable-generation · diffusion · discrete-representation · distributed-training · efficiency · embeddings · evaluation · fine-tuning · flow-matching · gan · guidance · image-generation · inference-optimization · instruction-tuning · interpretability · language-model · long-context · moe · multimodal · position-encoding · preference-optimization · pretraining · prompting · quantization · reasoning · reinforcement-learning · retrieval · rlhf · safety · sampling · scaling · science · search · self-improvement · self-play · self-supervised · state-space · synthetic-data · systems · test-time-compute · text-to-image · tool-use · transfer-learning · transformers · vae · video-generation · vision · world-models

agents¶

21 ReAct: Synergizing Reasoning and Acting in Language Models (2022)
24 Toolformer: Language Models Can Teach Themselves to Use Tools (2023)
30 Claude 3.5 Sonnet: Computer Use and Enhanced Capabilities (2024)
43 Claude 4 Family: The Agentic AI Leader (2025)
58 Generative Agents: Interactive Simulacra of Human Behavior (2023)
59 Model Context Protocol (MCP): An Open Standard for AI Tool Integration (2024)
62 AlphaEvolve: A Gemini-Powered Coding Agent for Designing Advanced Algorithms (2025)
78 Reflexion: Language Agents with Verbal Reinforcement Learning (2023)
84 SWE-bench: Can Language Models Resolve Real-World GitHub Issues? (2023)
100 Voyager: An Open-Ended Embodied Agent with Large Language Models (2023)
107 Human-Level Play in the Game of Diplomacy by Combining Language Models with Strategic Reasoning (CICERO) (2022)

alignment¶

05 Training Language Models to Follow Instructions with Human Feedback (InstructGPT) (2022)
14 Constitutional AI: Harmlessness from AI Feedback (2022)
17 LLaMA 2: Open Foundation and Fine-Tuned Chat Models (2023)
19 Direct Preference Optimization (DPO): Your Language Model is Secretly a Reward Model (2023)
38 GRPO: Group Relative Policy Optimization (2024)
51 Let's Verify Step by Step: Process Reward Models (2023)
63 Proximal Policy Optimization Algorithms (PPO) (2017)
79 Self-Instruct: Aligning Language Models with Self-Generated Instructions (2022)
83 Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training (2024)
85 Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena (2023)
96 Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations (2023)
103 KTO: Model Alignment as Prospect Theoretic Optimization (2024)

architecture¶

attention¶

01 Attention Is All You Need (2017)
16 FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness (2022)
54 RoFormer: Enhanced Transformer with Rotary Position Embedding (RoPE) (2021)
66 Neural Machine Translation by Jointly Learning to Align and Translate (Bahdanau Attention) (2014)
68 Highly Accurate Protein Structure Prediction with AlphaFold (AlphaFold 2) (2021)
75 GQA: Grouped-Query Attention (and Multi-Query Attention) (2023)
95 Mistral 7B (2023)

audio¶

40 GPT-4o: The First Omni Model (2024)
49 Whisper: Robust Speech Recognition via Large-Scale Weak Supervision (2022)

benchmarks¶

84 SWE-bench: Can Language Models Resolve Real-World GitHub Issues? (2023)
85 Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena (2023)

chain-of-thought¶

code¶

56 Codex: Evaluating Large Language Models Trained on Code (2021)
62 AlphaEvolve: A Gemini-Powered Coding Agent for Designing Advanced Algorithms (2025)
84 SWE-bench: Can Language Models Resolve Real-World GitHub Issues? (2023)
100 Voyager: An Open-Ended Embodied Agent with Large Language Models (2023)

computer-vision¶

73 Deep Residual Learning for Image Recognition (ResNet) (2015)
74 U-Net: Convolutional Networks for Biomedical Image Segmentation (2015)

controllable-generation¶

diffusion¶

06 Denoising Diffusion Probabilistic Models (DDPM) (2020)
07 High-Resolution Image Synthesis with Latent Diffusion Models (Stable Diffusion) (2021)
44 Sora and Diffusion Transformers (DiT): Video Generation as World Simulation (2022)
48 DALL-E 3: Improving Image Generation with Better Captions (2023)
69 Classifier-Free Diffusion Guidance (2021)
70 Denoising Diffusion Implicit Models (DDIM) (2020)
71 ControlNet: Adding Conditional Control to Text-to-Image Diffusion Models (2023)
72 Flow Matching and Rectified Flow: The New Default for Image Generation (Stable Diffusion 3) (2022)
74 U-Net: Convolutional Networks for Biomedical Image Segmentation (2015)
91 Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding (Imagen) (2022)
92 DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation (2022)
101 Accurate Structure Prediction of Biomolecular Interactions with AlphaFold 3 (2024)

discrete-representation¶

89 Neural Discrete Representation Learning (VQ-VAE) (2017)
90 Taming Transformers for High-Resolution Image Synthesis (VQ-GAN) (2020)

distributed-training¶

76 ZeRO and Megatron-LM: How Trillion-Parameter Models Are Actually Trained (2019)

efficiency¶

07 High-Resolution Image Synthesis with Latent Diffusion Models (Stable Diffusion) (2021)
10 LoRA: Low-Rank Adaptation of Large Language Models (2021)
16 FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness (2022)
20 Mamba: Linear-Time Sequence Modeling with Selective State Spaces (2023)
22 QLoRA: Efficient Finetuning of Quantized LLMs (2023)
27 DeepSeek-V3 Technical Report (2024)
33 LLaMA 3.3: Matching 405B Performance with 70B Parameters (2024)
37 Mixtral of Experts (and the Mixture-of-Experts Architecture) (2024)
45 Speculative Decoding: Fast Inference from Transformers (2022)
52 PagedAttention: Efficient LLM Serving with vLLM (2023)
75 GQA: Grouped-Query Attention (and Multi-Query Attention) (2023)
76 ZeRO and Megatron-LM: How Trillion-Parameter Models Are Actually Trained (2019)
86 GPTQ and AWQ: Post-Training Quantization for Large Language Models (2022)
95 Mistral 7B (2023)

embeddings¶

evaluation¶

81 Emergent Abilities of Large Language Models (and the Mirage Rebuttal) (2022)
84 SWE-bench: Can Language Models Resolve Real-World GitHub Issues? (2023)
85 Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena (2023)
96 Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations (2023)

fine-tuning¶

10 LoRA: Low-Rank Adaptation of Large Language Models (2021)
22 QLoRA: Efficient Finetuning of Quantized LLMs (2023)
71 ControlNet: Adding Conditional Control to Text-to-Image Diffusion Models (2023)
92 DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation (2022)

flow-matching¶

72 Flow Matching and Rectified Flow: The New Default for Image Generation (Stable Diffusion 3) (2022)

gan¶

02 Generative Adversarial Networks (GANs) (2014)
90 Taming Transformers for High-Resolution Image Synthesis (VQ-GAN) (2020)

guidance¶

69 Classifier-Free Diffusion Guidance (2021)

image-generation¶

02 Generative Adversarial Networks (GANs) (2014)
06 Denoising Diffusion Probabilistic Models (DDPM) (2020)
07 High-Resolution Image Synthesis with Latent Diffusion Models (Stable Diffusion) (2021)
48 DALL-E 3: Improving Image Generation with Better Captions (2023)
57 Auto-Encoding Variational Bayes (VAE) (2013)
69 Classifier-Free Diffusion Guidance (2021)
70 Denoising Diffusion Implicit Models (DDIM) (2020)
71 ControlNet: Adding Conditional Control to Text-to-Image Diffusion Models (2023)
72 Flow Matching and Rectified Flow: The New Default for Image Generation (Stable Diffusion 3) (2022)
89 Neural Discrete Representation Learning (VQ-VAE) (2017)
90 Taming Transformers for High-Resolution Image Synthesis (VQ-GAN) (2020)
91 Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding (Imagen) (2022)
92 DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation (2022)

inference-optimization¶

16 FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness (2022)
45 Speculative Decoding: Fast Inference from Transformers (2022)
52 PagedAttention: Efficient LLM Serving with vLLM (2023)
70 Denoising Diffusion Implicit Models (DDIM) (2020)
75 GQA: Grouped-Query Attention (and Multi-Query Attention) (2023)
86 GPTQ and AWQ: Post-Training Quantization for Large Language Models (2022)

instruction-tuning¶

05 Training Language Models to Follow Instructions with Human Feedback (InstructGPT) (2022)
46 LLaVA: Visual Instruction Tuning (2023)
79 Self-Instruct: Aligning Language Models with Self-Generated Instructions (2022)
80 FLAN: Finetuned Language Models Are Zero-Shot Learners (Instruction Tuning) (2021)

interpretability¶

language-model¶

03 BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding (2018)
04 Language Models are Few-Shot Learners (GPT-3) (2020)
15 LLaMA: Open and Efficient Foundation Language Models (2023)
17 LLaMA 2: Open Foundation and Fine-Tuned Chat Models (2023)
27 DeepSeek-V3 Technical Report (2024)
28 Qwen3: Technical Report (2025)
30 Claude 3.5 Sonnet: Computer Use and Enhanced Capabilities (2024)
33 LLaMA 3.3: Matching 405B Performance with 70B Parameters (2024)
36 GPT-4 Technical Report (2023)
41 Llama 4: Natively Multimodal Open-Source AI (2025)
42 GPT-5: Unified Intelligence (2025)
43 Claude 4 Family: The Agentic AI Leader (2025)
56 Codex: Evaluating Large Language Models Trained on Code (2021)
64 Language Models are Unsupervised Multitask Learners (GPT-2) (2019)
65 Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer (T5) (2019)
93 Improving Language Understanding by Generative Pre-Training (GPT-1) (2018)
94 PaLM: Scaling Language Modeling with Pathways (2022)
95 Mistral 7B (2023)
106 Evolutionary-Scale Prediction of Atomic-Level Protein Structure with a Language Model (ESM-2 / ESMFold) (2023)

long-context¶

moe¶

27 DeepSeek-V3 Technical Report (2024)
37 Mixtral of Experts (and the Mixture-of-Experts Architecture) (2024)
41 Llama 4: Natively Multimodal Open-Source AI (2025)
67 Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity (2021)

multimodal¶

08 Learning Transferable Visual Models From Natural Language Supervision (CLIP) (2021)
23 GPT-4V(ision): System Card (2023)
29 Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities (2025)
36 GPT-4 Technical Report (2023)
40 GPT-4o: The First Omni Model (2024)
41 Llama 4: Natively Multimodal Open-Source AI (2025)
46 LLaVA: Visual Instruction Tuning (2023)
47 Gemini 3: Google's Most Capable AI Model (2025)

position-encoding¶

54 RoFormer: Enhanced Transformer with Rotary Position Embedding (RoPE) (2021)

preference-optimization¶

pretraining¶

03 BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding (2018)
04 Language Models are Few-Shot Learners (GPT-3) (2020)
15 LLaMA: Open and Efficient Foundation Language Models (2023)
64 Language Models are Unsupervised Multitask Learners (GPT-2) (2019)
65 Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer (T5) (2019)
80 FLAN: Finetuned Language Models Are Zero-Shot Learners (Instruction Tuning) (2021)
88 Masked Autoencoders Are Scalable Vision Learners (MAE) (2021)
93 Improving Language Understanding by Generative Pre-Training (GPT-1) (2018)
94 PaLM: Scaling Language Modeling with Pathways (2022)
98 Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking (2024)

prompting¶

77 Self-Consistency Improves Chain of Thought Reasoning in Language Models (2022)
78 Reflexion: Language Agents with Verbal Reinforcement Learning (2023)
99 Self-Refine: Iterative Refinement with Self-Feedback (2023)

quantization¶

22 QLoRA: Efficient Finetuning of Quantized LLMs (2023)
86 GPTQ and AWQ: Post-Training Quantization for Large Language Models (2022)

reasoning¶

09 Chain-of-Thought Prompting Elicits Reasoning in Large Language Models (2022)
21 ReAct: Synergizing Reasoning and Acting in Language Models (2022)
25 Tree of Thoughts: Deliberate Problem Solving with Large Language Models (2023)
26 DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning (2025)
28 Qwen3: Technical Report (2025)
31 OpenAI o1: Learning to Reason with Reinforcement Learning (2024)
34 Meta Chain-of-Thought: Towards System 2 Reasoning in LLMs (2025)
35 rStar-Math: Small LLMs Can Master Math Reasoning with Self-Evolved Deep Thinking (2025)
39 RLVR: Reinforcement Learning from Verifiable Rewards (2024)
42 GPT-5: Unified Intelligence (2025)
50 Scaling LLM Test-Time Compute: The Theoretical Foundation for Reasoning Models (2024)
51 Let's Verify Step by Step: Process Reward Models (2023)
61 AlphaGeometry: Solving Olympiad Geometry Without Human Demonstrations (2024)
77 Self-Consistency Improves Chain of Thought Reasoning in Language Models (2022)
78 Reflexion: Language Agents with Verbal Reinforcement Learning (2023)
81 Emergent Abilities of Large Language Models (and the Mirage Rebuttal) (2022)
97 STaR: Bootstrapping Reasoning With Reasoning (Self-Taught Reasoner) (2022)
98 Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking (2024)
99 Self-Refine: Iterative Refinement with Self-Feedback (2023)
107 Human-Level Play in the Game of Diplomacy by Combining Language Models with Strategic Reasoning (CICERO) (2022)

reinforcement-learning¶

26 DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning (2025)
38 GRPO: Group Relative Policy Optimization (2024)
39 RLVR: Reinforcement Learning from Verifiable Rewards (2024)
63 Proximal Policy Optimization Algorithms (PPO) (2017)
102 Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm (2017)
105 Mastering Diverse Domains through World Models (DreamerV3) (2023)
107 Human-Level Play in the Game of Diplomacy by Combining Language Models with Strategic Reasoning (CICERO) (2022)

retrieval¶

rlhf¶

safety¶

sampling¶

70 Denoising Diffusion Implicit Models (DDIM) (2020)

scaling¶

04 Language Models are Few-Shot Learners (GPT-3) (2020)
12 Scaling Laws for Neural Language Models (2020)
18 Training Compute-Optimal Large Language Models (Chinchilla) (2022)
50 Scaling LLM Test-Time Compute: The Theoretical Foundation for Reasoning Models (2024)
64 Language Models are Unsupervised Multitask Learners (GPT-2) (2019)
67 Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity (2021)
76 ZeRO and Megatron-LM: How Trillion-Parameter Models Are Actually Trained (2019)
80 FLAN: Finetuned Language Models Are Zero-Shot Learners (Instruction Tuning) (2021)
81 Emergent Abilities of Large Language Models (and the Mirage Rebuttal) (2022)
94 PaLM: Scaling Language Modeling with Pathways (2022)

science¶

search¶

self-improvement¶

97 STaR: Bootstrapping Reasoning With Reasoning (Self-Taught Reasoner) (2022)
98 Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking (2024)
99 Self-Refine: Iterative Refinement with Self-Feedback (2023)
100 Voyager: An Open-Ended Embodied Agent with Large Language Models (2023)

self-play¶

102 Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm (2017)

self-supervised¶

88 Masked Autoencoders Are Scalable Vision Learners (MAE) (2021)
104 Genie: Generative Interactive Environments (2024)

state-space¶

20 Mamba: Linear-Time Sequence Modeling with Selective State Spaces (2023)

synthetic-data¶

systems¶

76 ZeRO and Megatron-LM: How Trillion-Parameter Models Are Actually Trained (2019)

test-time-compute¶

text-to-image¶

91 Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding (Imagen) (2022)

tool-use¶

21 ReAct: Synergizing Reasoning and Acting in Language Models (2022)
24 Toolformer: Language Models Can Teach Themselves to Use Tools (2023)
59 Model Context Protocol (MCP): An Open Standard for AI Tool Integration (2024)
78 Reflexion: Language Agents with Verbal Reinforcement Learning (2023)
100 Voyager: An Open-Ended Embodied Agent with Large Language Models (2023)

transfer-learning¶

93 Improving Language Understanding by Generative Pre-Training (GPT-1) (2018)

transformers¶

01 Attention Is All You Need (2017)
11 An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (Vision Transformer) (2020)
90 Taming Transformers for High-Resolution Image Synthesis (VQ-GAN) (2020)

vae¶

57 Auto-Encoding Variational Bayes (VAE) (2013)
89 Neural Discrete Representation Learning (VQ-VAE) (2017)

video-generation¶

44 Sora and Diffusion Transformers (DiT): Video Generation as World Simulation (2022)
104 Genie: Generative Interactive Environments (2024)

vision¶

08 Learning Transferable Visual Models From Natural Language Supervision (CLIP) (2021)
11 An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (Vision Transformer) (2020)
23 GPT-4V(ision): System Card (2023)
32 SAM 2: Segment Anything in Images and Videos (2024)
40 GPT-4o: The First Omni Model (2024)
46 LLaVA: Visual Instruction Tuning (2023)
73 Deep Residual Learning for Image Recognition (ResNet) (2015)
74 U-Net: Convolutional Networks for Biomedical Image Segmentation (2015)
88 Masked Autoencoders Are Scalable Vision Learners (MAE) (2021)

world-models¶

104 Genie: Generative Interactive Environments (2024)
105 Mastering Diverse Domains through World Models (DreamerV3) (2023)